SpaceX makes money from three business segments, and in 2025 only one of them turned a profit.
Starlink earned $4.4 billion from operations that year, while the rocket business lost $657 million and the AI business lost $6.4 billion.
For 24 years, those numbers were private.
Key Takeaways
SpaceX makes money from three segments — Connectivity (Starlink), Space, and AI — and in 2025 only Connectivity earned an operating profit.
Starlink made $4.4 billion of operating income on $11.4 billion of revenue, while Space lost $657 million and AI lost $6.4 billion.
SpaceX is EBITDA-positive and operating-loss negative at the same time: $6.6 billion of Adjusted EBITDA against a $2.6 billion operating loss on $18.7 billion of revenue.
The rocket segment lost money because it absorbed $3.0 billion of Starship development and books no revenue for the 123 Starlink launches it flew.
Government work is contract revenue that SpaceX competes for and gets paid on delivery, which is not the same thing as a grant.
These are IPO-filing figures, not an earnings track record — SpaceX's first quarterly report as a public company will be its first ever.
SpaceX splits its business into three reporting segments: Space, Connectivity, and AI.
That split matters more than the rocket-versus-satellite framing most people use, because the three segments do not resemble each other financially at all.
One of them earns money. The other two spend it.
Segment | What's in it | 2025 revenue | 2025 income (loss) from operations | Share of revenue |
Connectivity | Starlink broadband, enterprise, government, mobile | $11,387M | +$4,423M | 61% |
Space | Falcon 9, Falcon Heavy, Dragon, Starship | $4,086M | $(657)M | 22% |
AI | xAI, Grok, X, compute infrastructure | $3,201M | $(6,355)M | 17% |
Consolidated |
| $18,674M | $(2,589)M | 100% |
The segment numbers add up cleanly, which is what makes them useful.
Add Connectivity's $4,423 million profit to Space's $657 million loss and AI's $6,355 million loss, and you land exactly on the company's $2,589 million consolidated operating loss.
That arithmetic is the honest answer to how SpaceX makes money: Starlink makes it, and the other two segments spend it faster than Starlink earns it.
The money follows the subscribers.
That gap — profit growing faster than revenue — is what a subscription business looks like when it works.
The satellites are the cost, and they are already up there.
That is why Connectivity's operating income grew 120.4% in 2025 on 49.8% revenue growth — profit rising more than twice as fast as the top line.
There is a nuance the growth numbers hide: average revenue per subscriber has been falling.
SpaceX attributes that to subscriber mix — growth is coming from international markets that pay less, which drags the average down even when nobody's bill changes.
Whether that mix keeps diluting the average as Starlink expands is the open question in its economics.
The Space segment brought in $4,086 million for that work and lost $657 million.
The loss is not a sign of a broken business — it is a choice.
The segment spent $3,004 million developing Starship in 2025, and that spending sits inside the same segment as the Falcon 9 revenue.
Strip Starship out and the picture flips; leave it in, and the rocket business is a research program with a launch service attached.
The other thing the launch count hides is who the launches are for.
There is a structural reason those flights don't help the Space segment.
SpaceX's own filing states that Space revenue reflects external customer launches only, with no inter-segment revenue for launches that deploy its own satellites.
So the 123 Starlink flights earned that segment nothing — the cost sits in Space, and the benefit shows up in Connectivity.
Falcon 9's largest customer is SpaceX, which means the launch business is mostly the delivery mechanism for the subscription business rather than a standalone revenue engine.
SpaceX bought xAI in February 2026, and with it came Grok, the X platform, and the largest AI training clusters the company says exist on Earth.
The AI segment generated $3,201 million in revenue in 2025 and lost $6,355 million from operations.
It also consumed $12,727 million in capital expenditure that year, against $4,178 million for Connectivity and $3,832 million for Space.
SpaceX put more capital into AI infrastructure in 2025 than into rockets and satellites put together.
Read the segment table again with that in mind: without AI, SpaceX's remaining two segments earned $3,766 million from operations in 2025 instead of losing $2,589 million.
The AI segment is what turned a profitable space-and-internet company into a loss-making one, and SpaceX says that is deliberate.
There is one wrinkle in how these numbers are presented.
Because the xAI deal joined two companies already under the same owner, accounting rules required SpaceX to restate its earlier financials as if it had always owned xAI — so the 2025 AI figures describe a business SpaceX did not actually own during 2025.
The revenue and the losses are real; the ownership in that year is a presentation convention, and it is worth knowing before you compare 2025 to anything older.
The honest answer is that SpaceX is profitable and unprofitable at the same time, depending on which line you read.
One is a GAAP figure and one is a measure SpaceX defines itself — the gap between them is the most useful thing in the filing.
Measure | 2025 | What it tells you |
Revenue | $18,674M | The business is large and growing |
Adjusted EBITDA | +$6,584M | The core operations throw off cash |
Loss from operations | $(2,589)M | After the real costs of running it, it loses money |
Starlink alone (operating income) | +$4,423M | One segment carries the company |
Starship R&D (inside Space) | $3,004M | Where the rocket segment's profit went |
Total capital expenditure | | What SpaceX spent building the future |
Adjusted EBITDA is a non-GAAP measure defined by SpaceX in its Form S-1. Year ended December 31, 2025.
Adjusted EBITDA strips out several real costs, and the biggest one is depreciation.
SpaceX spent $20,737 million on physical assets in 2025 — satellites, rockets, data centers, ground stations.
Those assets wear out, and accounting rules make the company write down a slice of their value every year for as long as they last.
Adjusted EBITDA ignores that write-down. Loss from operations does not.
This is why capital expenditure does not directly cause an operating loss, even though people often say it does: the cash goes out in one year, but it hits the income statement slowly, over the whole life of the equipment.
The two numbers answer two different questions.
Adjusted EBITDA of $6,584 million says the day-to-day business generates cash.
It does not say the business funds itself — SpaceX spent $20,737 million on capital projects that same year, which is why it carries debt and raised $75 billion in its IPO.
A $2,589 million operating loss says that once you count what it costs to replace everything SpaceX is wearing out and to fund everything it is inventing, the company is spending more than it earns — on purpose, and for now.
A persistent misreading of SpaceX's business treats government contracts as government funding.
They are not the same thing, and the difference changes how you should read the risk.
The government money in SpaceX's Space segment is contract revenue, not grant funding.
The scale of that work is easy to state without guesswork, because SpaceX states it.
In 2025 the company launched 11 of the 12 National Security Space Launch medium and heavy lift missions and all five U.S. crew and cargo missions to the International Space Station for NASA, making it the primary launch provider for the U.S. government.
Government revenue earned by competitive bid carries a different risk profile than a subsidy: it has to be re-won each cycle, and federal contracts can be modified or terminated by the government that awarded them.
Before June 12, 2026, most people could not buy SpaceX stock, but you could buy exposure to it — and the price of that exposure was answering a different question than the stock price does now.
MEXC ran both sides of that transition, through a SPACEX(PRE) Launchpad and SPCX futures before the listing, SPCX equity after it, which makes the mechanism visible from the inside.
A pre-IPO instrument prices scarcity of access.
The allocation is fixed and the queue is not, so demand does not move the subscription price — it moves how little of it you get.
An IPO prices something else entirely.
MEXC's SPCX futures were built to roll over automatically at listing so that a position taken before the IPO tracks the real stock price afterward — and that roll-over is the exact moment the access premium stops mattering and the filings start.
This is why pre-IPO enthusiasm is not a forecast: the two markets are pricing two different scarcities, and only one of them is scarcity of profit.
The question the market asks about SpaceX changed on June 12, and the segment table is where the new answer lives.
How does SpaceX make money?
SpaceX earns revenue from three segments — Connectivity (Starlink subscriptions), Space (launches for outside customers), and AI (xAI, Grok, and X) — with Starlink contributing 61% of 2025 revenue and all of the operating profit.
Is SpaceX profitable?
It depends on the measure: SpaceX generated $6.6 billion in Adjusted EBITDA in 2025 but lost $2.6 billion from operations, because Adjusted EBITDA excludes $6.7 billion of depreciation and amortization and $1.9 billion of share-based compensation.
What is SpaceX's annual revenue?
SpaceX reported $18.7 billion in revenue for 2025 in its Form S-1, its first publicly filed financial statements in 24 years.
How much of SpaceX's revenue comes from the government?
SpaceX does not break out a government revenue line, and government work runs across all three segments — launches in Space, Starshield and Government Solutions in Connectivity, and xAI Gov in AI, and the company launched 11 of 12 National Security Space Launch missions and all five NASA ISS crew and cargo missions in 2025.
Does SpaceX get government funding?
The government payments in SpaceX's launch business are contract revenue — SpaceX competes for the award and gets paid when it delivers a launch, not for existing.
How much money does SpaceX make per launch?
Dividing segment revenue by launch count is misleading, because 123 of SpaceX's 165 Falcon 9 flights in 2025 carried its own Starlink satellites — and SpaceX books no revenue for those.
Why did SpaceX report a loss if Starlink is profitable?
Starlink's $4.4 billion operating profit was outweighed by a $657 million loss in the Space segment, which absorbed $3.0 billion of Starship development, and a $6.4 billion loss in the newly acquired AI segment.
Where can I find SpaceX's official financial results?
The public story about SpaceX is rockets. The filings say the rockets are the delivery system, and Starlink is the business.
Starlink earned $4.4 billion from operations in 2025 while the launch business lost $657 million and the AI business lost $6.4 billion — and three out of every four Falcon 9 flights that year carried Starlink's own satellites rather than a customer's.
That is not a criticism of SpaceX. It is a description of a company that built the cheapest road to orbit and then decided the most valuable thing to put on it was its own cargo.
Whether the AI segment eventually earns back what it is spending is the open question in SPCX, and it is a question the filings will answer four times a year now rather than never.
For anyone deciding what to do with that, the segment table above is a better starting point than the launch count.